On 1/27/2014 4:12 PM, usher99@xxxxxxx wrote:
> LTM writes:
>
>>> I understand our culture's obsession with measurement, rankings and
>>> winners. But when the rankings do not reflect actual performance
>>> differences, I can't see how they are of use. What the H are they
>>> measuring? Why?
> One has to dig but they do explain the measurments. Adorama ranked
> them as if the exact order mattered and did not take into account if
> the differences were significant from a practical vantage point.
Agreed, and that's exactly one of the reasons I object to DXOMark's rating
system. We all know the ASA/ISO sensor speed
numbering system is crazy. As long, term users of that system, we tend to get
fooled less than those not used to it. But
let's face it, 25,600 vs. 3,200 SOUNDs WAY faster than f2 vs. f5.6. Had the ISO
gone with something more like the DIN
value system, would we think about 'speed' somewhat differently?
So using a measure like ISO numbers for noise is just asking for it to be
misunderstood by fools who just read the
numbers. That's why I think some of their measures are dumb.
The second reason I ask what are they measuring is a broader question. I could
put it differently, "How does what they
are measuring relate to what we see in an image?"
> http://www.dxomark.com/en/About/Sensor-scores/Use-Case-Scores
>
> Appears a 25% difference in ISO score is about 1/3 EV and only
> "slightly noticeable".
>
> Curiously the best optics according to DXO perform much better n the
> EM-1 than the Gx-7--perhaps due to the more robust AA filter.
>
> http://www.dxomark.com/Reviews/Best-lenses-for-the-16-MPix-Panasonic-Lumix-DMC-GX7-Part-1/How-does-sharpness-compare-between-the-Panasonic-Lumix-DMC-GX7-and-the-Olympus-OM-D-E-M1
>
> The difference is most marked for the best performing optic the 75/1.8
> and less of difference with other lenses according to DXO's "Perceputal MP"
> measure of sharpness,
>
> http://www.dxomark.com/enà¹%20ฟà¸%20à¹%20ฟà¸%20à¹%20ฟà¸%20/Reviews/Looking-for-new-photo-gear-DxOMark-s-Perceptual-Megapixel-can-help-you
>
> Wonder what Moose thinks of this?
I think it's far more complex than we might wish it were. I invite you to go to
this page. Set the cameras to E-M1,
E-M5, Panny GX7 and Fuji X-Pro 1 and ISO to 200.
I don't know how other people's eyes see this. I can only say what I see. At
first glance, it may appear that the M1 is
sharper (has more visible detail). M5 looks less sharp, but cleaner. GX7 just
looks soft. The X-Pro 1 has a very clean
look, with almost 3D modeling.
With closer attention, the E1 looks to me a lot like an image that has been
slightly over processed. There's a slightly
odd edge effect. Look at the line of forehead against darker background; isn't
that a halo? Looks too like higher
contrast has lost some shadow detail.
The E5 has softer edges and is less contrasty, but is there actually less
detail, or is it less emphasized?
I don't know what's up with the ISO 200 GX7 image. It doesn't track with other
comparisons in the view. May be an
anomalous error.
The X-Pro 1 looks very attractive at first, but look at the lower line of
beading on the crown. The beads blur together,
losing the clear distinctions seen with the other cameras.
Now, go to ISO 400. Three remain much the same, but the GX7 jumps up to show
more, clearer detail.
ISO 800, and the M5 seems to have picked up detail. The M1 now looks a bit more
over processed.
ISO 1600 & 3200, M1 starts to look pretty unnatural, M5 is arguably the best
looking, overall. The Fuji is remarkably clean.
ISO 3200, the µ4/3 cameras fall apart. The Fuji is fabulous, by comparison, and
has lost remarkably little from lower ISOs
Thoughts:
--------------------------
How can a single number reflect anything of this variation?
The lack of an AA filter both gives and takes away. Whether caused by the same
physics/optics phenomena as moire, it can
give some unattractive results at the pixel level.
Now to the next level:
--------------------------
I'm not just interested in what the image out of camera looks like. I'm
interested in how it looks after post. Consider
this comparison, with bits of NR, deconvolution, LCE and Curves applied to the
M5 ISO 200 image.
<http://www.moosemystic.net/Gallery/tech/E-M5/E-M1ewtal.htm>
Seems to me it is comparable in visual sharpness and detail to the M1, a bit
cleaner and slightly more visually
attractive. Does the lack of an AA filter actually improve the images?
Now, what if I do that to the M1 image? I've been at this sort of post for
quite a while, processing thousands of
images, and I'm not going to waste my time on that. Images with those
characteristics just don't get better. Maybe if I
apply my own "AA filter" with blur, then process further? But which kind of
blur, and how much? Life is too short.
More thoughts:
--------------------------
I wish I could take you on a tour of the test image on page 17, but I don't
have the time and energy. I've spent some
time with it, comparing E-M1, E-M5 and GX7. I've found specific subjects and
ISOs where each one outperforms the others.
Contrary to what one might expect, the M1 clearly out resolves the other two at
ISO 200 and low detail. Look at the
cards, upper right, and drawing of guy with beard, right centerish. Cards
clearly sharper on M1, fine line detail
virtually the same, would be just the same after post. By ISO 400, maybe slight
advantage M5, ISO 800, M5 clearly
sharper. They'd all end up the same after post.
I'm not convinced that the lack of an AA filter in the E1 is an advantage,
overall, for actual image making. For
practical use, on the part of this one test/criterion I've shown/discussed,
it's a draw to me. In other looking and
testing, the GX7 holds detail in shadow, responds to NR there better and has
significantly fewer wormy artifacts.
Close Attention Moose
--
What if the Hokey Pokey *IS* what it's all about?
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|